On Thu, Jan 09, 2025 at 03:48:56PM +0100, Ard Biesheuvel wrote:
> On Tue, 7 Jan 2025 at 13:05, Breno Leitao <leitao@xxxxxxxxxx> wrote:
> > On Tue, Jan 07, 2025 at 12:24:03PM +0100, Ard Biesheuvel wrote:
> > We are seeing a bunch of `Corrupted EFI Memory Attributes Table
> > detected!` in the Meta fleet, and this is something we are
> > investigating.
> >
> > We highly think this is related to some kexec overwrites, and when we
> > get here, the EFI table is completely garbage. I haven't seen this
> > problem on cold boot.
> >
>
> It likely means the memory is not reserved correctly.
>
> Could you check whether this
>
> --- a/drivers/firmware/efi/memattr.c
> +++ b/drivers/firmware/efi/memattr.c
> @@ -56,7 +56,7 @@ int __init efi_memattr_init(void)
> }
>
> tbl_size = sizeof(*tbl) + size;
> - memblock_reserve(efi_mem_attr_table, tbl_size);
> + efi_mem_reserve(efi_mem_attr_table, tbl_size);
> set_bit(EFI_MEM_ATTR, &efi.flags);
>
> unmap:
>
>
> makes any difference?
It doesn't seem so. This is how I tested it.
I've cherry-picked it on top of Linus' master eea6e4b4dfb8 ("Merge tag
'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi"),
and booted a system with this kernel.
